upk
This commit is contained in:
nose
@@ -19,6 +19,8 @@ from datetime import datetime
|
||||
from pathlib import Path
|
||||
from typing import Optional, Dict, Any, List, Tuple, Set
|
||||
|
||||
from .utils import sha256_file
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
# Try to import optional dependencies
|
||||
@@ -455,6 +457,18 @@ class LocalLibraryDB:
|
||||
logger.error(f"[get_or_create_file_entry] ❌ Error getting/creating file entry for {file_path}: {e}", exc_info=True)
|
||||
raise
|
||||
|
||||
def get_file_id(self, file_path: Path) -> Optional[int]:
|
||||
"""Get the file ID for a file path, or None if not found."""
|
||||
try:
|
||||
str_path = str(file_path.resolve())
|
||||
cursor = self.connection.cursor()
|
||||
cursor.execute("SELECT id FROM files WHERE file_path = ?", (str_path,))
|
||||
row = cursor.fetchone()
|
||||
return row[0] if row else None
|
||||
except Exception as e:
|
||||
logger.error(f"Error getting file ID for {file_path}: {e}", exc_info=True)
|
||||
return None
|
||||
|
||||
def get_metadata(self, file_path: Path) -> Optional[Dict[str, Any]]:
|
||||
"""Get metadata for a file."""
|
||||
try:
|
||||
@@ -748,6 +762,177 @@ class LocalLibraryDB:
|
||||
logger.error(f"Error removing tags for {file_path}: {e}", exc_info=True)
|
||||
raise
|
||||
|
||||
def set_relationship(self, file_path: Path, related_file_path: Path, rel_type: str = "alt") -> None:
|
||||
"""Set a relationship between two local files.
|
||||
|
||||
Args:
|
||||
file_path: Path to the file being related
|
||||
related_file_path: Path to the related file
|
||||
rel_type: Type of relationship ('king', 'alt', 'related')
|
||||
"""
|
||||
try:
|
||||
str_path = str(file_path.resolve())
|
||||
str_related_path = str(related_file_path.resolve())
|
||||
|
||||
file_id = self.get_or_create_file_entry(file_path)
|
||||
related_file_id = self.get_or_create_file_entry(related_file_path)
|
||||
|
||||
cursor = self.connection.cursor()
|
||||
|
||||
# Get hashes for both files
|
||||
file_hash = sha256_file(file_path)
|
||||
related_file_hash = sha256_file(related_file_path)
|
||||
|
||||
if not file_hash or not related_file_hash:
|
||||
logger.warning(f"Cannot set relationship: missing hash for {file_path} or {related_file_path}")
|
||||
return
|
||||
|
||||
# Store the hashes in the files table for future lookups
|
||||
cursor.execute("""
|
||||
UPDATE files SET file_hash = ? WHERE id = ?
|
||||
""", (file_hash, file_id))
|
||||
cursor.execute("""
|
||||
UPDATE files SET file_hash = ? WHERE id = ?
|
||||
""", (related_file_hash, related_file_id))
|
||||
|
||||
# Get current relationships
|
||||
cursor.execute("""
|
||||
SELECT relationships FROM metadata WHERE file_id = ?
|
||||
""", (file_id,))
|
||||
|
||||
row = cursor.fetchone()
|
||||
# Use index access to be safe regardless of row_factory
|
||||
relationships_str = row[0] if row else None
|
||||
|
||||
try:
|
||||
if relationships_str:
|
||||
relationships = json.loads(relationships_str)
|
||||
else:
|
||||
relationships = {}
|
||||
except (json.JSONDecodeError, TypeError):
|
||||
relationships = {}
|
||||
|
||||
# Ensure relationships is a dict (handle case where DB has a list)
|
||||
if not isinstance(relationships, dict):
|
||||
relationships = {}
|
||||
|
||||
# Ensure rel_type key exists
|
||||
if rel_type not in relationships:
|
||||
relationships[rel_type] = []
|
||||
|
||||
# Add the relationship (store as hash string)
|
||||
if related_file_hash not in relationships[rel_type]:
|
||||
relationships[rel_type].append(related_file_hash)
|
||||
|
||||
# Save the updated relationships for the main file
|
||||
cursor.execute("""
|
||||
INSERT INTO metadata (file_id, relationships)
|
||||
VALUES (?, ?)
|
||||
ON CONFLICT(file_id) DO UPDATE SET
|
||||
relationships = excluded.relationships,
|
||||
time_modified = CURRENT_TIMESTAMP
|
||||
""", (file_id, json.dumps(relationships)))
|
||||
|
||||
logger.debug(f"Set {rel_type} relationship: {str_path} ({file_hash}) -> {str_related_path} ({related_file_hash})")
|
||||
|
||||
# Set reverse relationship (bidirectional)
|
||||
# For 'alt' and 'related', the reverse is the same
|
||||
# For 'king', the reverse is 'subject' (or we just use 'alt' for simplicity as Hydrus does)
|
||||
# Let's use the same type for now to keep it simple and consistent with Hydrus 'alternates'
|
||||
reverse_type = rel_type
|
||||
|
||||
# Update the related file
|
||||
cursor.execute("""
|
||||
SELECT relationships FROM metadata WHERE file_id = ?
|
||||
""", (related_file_id,))
|
||||
|
||||
row = cursor.fetchone()
|
||||
relationships_str = row[0] if row else None
|
||||
|
||||
try:
|
||||
if relationships_str:
|
||||
reverse_relationships = json.loads(relationships_str)
|
||||
else:
|
||||
reverse_relationships = {}
|
||||
except (json.JSONDecodeError, TypeError):
|
||||
reverse_relationships = {}
|
||||
|
||||
if not isinstance(reverse_relationships, dict):
|
||||
reverse_relationships = {}
|
||||
|
||||
if reverse_type not in reverse_relationships:
|
||||
reverse_relationships[reverse_type] = []
|
||||
|
||||
if file_hash not in reverse_relationships[reverse_type]:
|
||||
reverse_relationships[reverse_type].append(file_hash)
|
||||
|
||||
# Save the updated reverse relationships
|
||||
cursor.execute("""
|
||||
INSERT INTO metadata (file_id, relationships)
|
||||
VALUES (?, ?)
|
||||
ON CONFLICT(file_id) DO UPDATE SET
|
||||
relationships = excluded.relationships,
|
||||
time_modified = CURRENT_TIMESTAMP
|
||||
""", (related_file_id, json.dumps(reverse_relationships)))
|
||||
|
||||
self.connection.commit()
|
||||
|
||||
except Exception as e:
|
||||
logger.error(f"Error setting relationship: {e}", exc_info=True)
|
||||
raise
|
||||
|
||||
def find_files_pointing_to(self, target_path: Path) -> List[Dict[str, Any]]:
|
||||
"""Find all files that have a relationship pointing to the target path.
|
||||
|
||||
Args:
|
||||
target_path: The file path to look for in other files' relationships
|
||||
|
||||
Returns:
|
||||
List of dicts with {path, type} for files pointing to target
|
||||
"""
|
||||
try:
|
||||
# Get the hash of the target file
|
||||
target_hash = sha256_file(target_path)
|
||||
if not target_hash:
|
||||
logger.warning(f"Cannot find files pointing to {target_path}: unable to compute hash")
|
||||
return []
|
||||
|
||||
cursor = self.connection.cursor()
|
||||
|
||||
# Scan all metadata (this might be slow on huge DBs but fine for local library)
|
||||
# We select file_path and relationships json
|
||||
cursor.execute("""
|
||||
SELECT f.file_path, m.relationships
|
||||
FROM metadata m
|
||||
JOIN files f ON m.file_id = f.id
|
||||
WHERE m.relationships LIKE ?
|
||||
""", (f"%{target_hash}%",))
|
||||
|
||||
results = []
|
||||
|
||||
for row in cursor.fetchall():
|
||||
f_path = row[0]
|
||||
rels_json = row[1]
|
||||
|
||||
try:
|
||||
rels = json.loads(rels_json)
|
||||
if isinstance(rels, dict):
|
||||
for r_type, hashes in rels.items():
|
||||
if isinstance(hashes, list):
|
||||
# Check if target hash is in this relationship type
|
||||
if target_hash in hashes:
|
||||
results.append({
|
||||
"path": f_path,
|
||||
"type": r_type
|
||||
})
|
||||
except (json.JSONDecodeError, TypeError):
|
||||
continue
|
||||
|
||||
return results
|
||||
except Exception as e:
|
||||
logger.error(f"Error finding files pointing to {target_path}: {e}", exc_info=True)
|
||||
return []
|
||||
|
||||
def get_note(self, file_path: Path) -> Optional[str]:
|
||||
"""Get note for a file."""
|
||||
try:
|
||||
@@ -1076,6 +1261,11 @@ class LocalLibraryDB:
|
||||
if not text:
|
||||
return True
|
||||
try:
|
||||
# Check if connection is valid
|
||||
if not self.connection:
|
||||
logger.warning(f"Database connection not available for worker {worker_id}")
|
||||
return False
|
||||
|
||||
cursor = self.connection.cursor()
|
||||
cursor.execute("SELECT stdout FROM worker WHERE worker_id = ?", (worker_id,))
|
||||
row = cursor.fetchone()
|
||||
@@ -1097,6 +1287,13 @@ class LocalLibraryDB:
|
||||
|
||||
self.connection.commit()
|
||||
return cursor.rowcount > 0
|
||||
except sqlite3.ProgrammingError as e:
|
||||
# Handle "Cannot operate on a closed database" gracefully
|
||||
if "closed database" in str(e).lower():
|
||||
logger.warning(f"Database connection closed, cannot append stdout for worker {worker_id}")
|
||||
return False
|
||||
logger.error(f"Error appending stdout to worker {worker_id}: {e}", exc_info=True)
|
||||
return False
|
||||
except Exception as e:
|
||||
logger.error(f"Error appending stdout to worker {worker_id}: {e}", exc_info=True)
|
||||
return False
|
||||
@@ -1599,3 +1796,23 @@ class LocalLibrarySearchOptimizer:
|
||||
if not self.db:
|
||||
return None
|
||||
return self.db.search_by_hash(file_hash)
|
||||
|
||||
def set_relationship(self, file_path: Path, related_file_path: Path, rel_type: str = "alt") -> None:
|
||||
"""Set a relationship between two files in the database.
|
||||
|
||||
Delegates to LocalLibraryDB.set_relationship().
|
||||
|
||||
Args:
|
||||
file_path: Path to the first file
|
||||
related_file_path: Path to the related file
|
||||
rel_type: Type of relationship ('king', 'alt', 'related', etc.)
|
||||
"""
|
||||
if not self.db:
|
||||
return
|
||||
self.db.set_relationship(file_path, related_file_path, rel_type)
|
||||
|
||||
def find_files_pointing_to(self, target_path: Path) -> List[Dict[str, Any]]:
|
||||
"""Find all files that have a relationship pointing to the target path."""
|
||||
if not self.db:
|
||||
return []
|
||||
return self.db.find_files_pointing_to(target_path)
|
||||
|
||||
Reference in New Issue
Block a user